Writing study: MARCO: Navigating the Unseen Space of Semantic Correspondence

CVPR overviewCorpus atlasBy contribution typeVisual galleryRules & templateWriting patternsPaper studies

← CVPR 2026 oral writing corpus

Claudia Cuttano, Gabriele Trivigno, Carlo Masone, Stefan Roth · Official CVPR 2026 oral record

Observed writing choices, not acceptance rules. This study examines the inspected manuscript version. It does not establish why the paper was accepted or that its presentation choices should be copied unchanged.

Figure, table and section ledger

Editorial contribution type: Training, objectives & optimization (primary).

Layout extraction, not a required paper template. Counts are distinct caption labels; a multi-panel figure counts once. Section spans are estimated from heading positions, including floats and whitespace. Method and limitations.

Canonical PDF

10 physical pages

Before references / appendix

≈ 8.17 page-equivalents

Main figures / tables

5 / 4

Attached appendix figures / tables

0 / 0

Attached appendix span

Separate file recorded below; not in this PDF total

Before references / appendix: ≈ 81.7%References / intervening material: ≈ 18.3%Estimated layout envelopes, not text/artwork area. Unresolved appendix boundaries remain within the undivided post-body material.

Extraction notes: No attached appendix boundary detected. Zero means no recognized caption, not independently verified absence.

Section-by-section lengths and source pages

The title/abstract/front matter precede the first numbered heading. A span ends at the next detected top-level heading, not at the last sentence; end-page whitespace is included. References and statements are listed separately. These are not measurements of text-only or artwork area.

Top-level heading	Starts on PDF page	Estimated span	Share of entire PDF
1. Introduction	1	≈ 1.27 pp.	12.7%
2. Related Work	3	≈ 0.62 pp.	6.2%
3. Semantic Correspondence with MARCO	3	≈ 2.88 pp.	28.8%
4. Experiments	6	≈ 2.28 pp.	22.8%
5. Conclusion	8	≈ 0.34 pp.	3.4%
References	9	≈ 1.83 pp.	18.3%

All detected figure and table locations

Main text

• Figure 1 · p. 1
• Figure 2 · p. 2
• Figure 3 · p. 4
• Figure 4 · p. 6
• Figure 5 · p. 6
• Table 1 · p. 7
• Table 2 · p. 7
• Table 3 · p. 8
• Table 4 · p. 8

Attached appendix

No recognized caption labels.

Separate source files—excluded from canonical-PDF totals

A separate author version may repeat main text and references. Whole-file lengths and counts are not extra appendix material and are never added to the canonical PDF totals. Read the study's source scope for which sections received substantive reading versus layout inspection.

Additional source 1

Separate PDF (or ZIP containing it): whole-file layout of 8 physical pages; 2 detected figures and 12 tables. These may include overlapping main-text material.

Located appendix only: starts on physical p.1; ≈ 7.94 page-equivalents, 2 figure labels and 12 table labels. This interval is distinct from the whole file.

Detected heading	PDF page	Estimated span
Supplementary Material	1	≈ 0.23 pp.
A. Broad Applicability of our Dense Self-	1	≈ 0.56 pp.
B. Pre-training on AP-10k	1	≈ 0.33 pp.
C. Additional Ablations	2	≈ 1.24 pp.
D. Use of Object Masks	3	≈ 0.38 pp.
E. Computational Cost	3	≈ 0.37 pp.
F. Pseudocode of Dense Self-Distillation	4	≈ 0.13 pp.
G. Details on the MP-100 Benchmark	4	≈ 2.59 pp.
H. Qualitative Examples	6	≈ 1.82 pp.
References	8	≈ 0.28 pp.

• Table 5 · p. 1
• Table 6 · p. 2
• Table 7 · p. 2
• Table 8 · p. 3
• Table 9 · p. 3
• Table 10 · p. 3
• Table 11 · p. 3
• Table 12 · p. 4
• Table 13 · p. 5
• Figure 6 · p. 5
• Table 14 · p. 7
• Figure 7 · p. 7
• Table 15 · p. 8
• Table 16 · p. 8

Argument and organization

1. Method §§3.1–3.3

   Separate compact architecture, precise supervised localization and dense flow-anchored self-distillation.

2. Experiments §§4.1–4.3

   Test standard matching first, unseen keypoints/categories second, then isolate the two goals in ablations.

Problem, claim and contribution

The paper defines a gap between benchmark landmarks and arbitrary user queries, motivating unseen-keypoint evaluation rather than only improving the familiar PCK leaderboard.

PDF pp. 1–2, Figures 1–2 · Main paper · Additional source 2

Argument and section order

Precision and generalization are consistently paired from the introduction through distinct objectives and the two datasets used for component analysis.

§§3.2–3.3, PDF pp. 4–6; §4.3, PDF p. 8 · Main paper · Additional source 2

Figures, captions and visual explanation

The flow-anchoring explanation shows source/target images, displacement clusters, kept and discarded regions and final correspondences; an incorrect coherent cluster is explicitly illustrated.

Figure 4, PDF p. 6 · Main paper · Additional source 2

Tables, comparisons and evidence

The standard-results caption marks depth and mask supervision, while separate tables evaluate unseen categories and landmarks. The ablation caption states that component groups are independently assessed with the others fixed.

Tables 1–4, PDF pp. 7–8 · Main paper · Additional source 2

Appendix and reproducibility

The supplement tests transplanting the loss to Geo-SC, controls extra pretraining, estimates pseudo-label noise sensitivity and gives pseudocode plus a matched inference-cost protocol.

Supplement §§A–H, PDF pp. 1–8; Tables 5–12 and Algorithm 1 · Main paper · Additional source 2

Clarity and writing conventions

The problem statement defines descriptors, lattice and correspondence sets before losses, then the discussion distinguishes plausible local motion from globally correct matching.

PDF p. 3, problem statement; PDF pp. 5–6, flow anchoring · Main paper · Additional source 2

What to reuse—and what not to infer

Transferable lesson: If generalization is the contribution, define the unseen dimension and reserve separate result tables for it; do not let in-domain accuracy stand in for transfer.

Boundary: Training masks, depth and extra pretraining differ across methods; preserve the table footnotes and the distinction between seen categories with new landmarks and genuinely unseen categories.

Apply this lesson using the argument and evidence worksheet.

Inspection record and sources

Checked

2026-09-26

Reading status

Writing-focused manuscript scan + selected visuals inspected

Main text

Manuscript-wide writing scan of PDF pp. 1–10: framing, argument sequence, section organization, figure/table captions, evaluation interpretation and limitations; reference sections identified. This is a writing-focused scan, not an exhaustive technical proof or reproduction audit. Main PDF pages 3 and 6 were rendered and visually inspected.

Appendix / supplement

Writing-focused scan of the separate supplementary PDF, pp. 1–8, including its organization, captions, protocol/evidence detail and limitation material where present; proofs and code listings were inspected for exposition, not independently verified or executed. Supplement §§A–H, PDF pp. 1–8; Tables 5–12 and Algorithm 1. Linked video/audio demonstrations and non-PDF archive contents were not assessed.

Visually inspected PDF pages

3, 6

This is a writing-focused scan of the manuscript and the stated supplementary scope: framing, organization, visual explanation, evidence presentation and wording. It is not a line-by-line proof audit, reproduction of results or exhaustive scientific review.

• Official CVPR 2026 CVF open-access accepted manuscript; retrieved 2026-09-26 · 10 pages
• Official CVPR 2026 CVF supplementary PDF; retrieved 2026-09-26 · 8 pages